Since the implementation of registration system reform, the information disclosure mechanisms of Chinese listed companies have consistently garnered attention. The purpose of information disclosure is to effectively transfer information from within an organization to external parties, with public texts serving a crucial role in this process. However, the conciseness and accuracy of this information largely depend on the willingness and strategy of the discloser. Based on this, this paper aims to construct a systematic readability assessment system using linguistics theories, comprising 14 sub-indicators from four dimensions—Morphology, Grammar, Semantics, and Multimodal—to measure the comprehensive readability level in the annual reports of listed companies through a hierarchical dimensionality reduction method. Furthermore, this paper explores the manipulative information disclosure strategies of managers driven by self-serving incentives. This paper selected more than 4000 listed companies as the sample for its empirical analysis. It is revealed that companies with poor operating results in the current period tend to have low readability levels in their annual reports. The readability level in the current period has a positive impact on the stock market in the next one to three years, especially for companies with stock returns below the industry average. Moreover, companies with lower readability levels in their annual reports tend to demonstrate higher excess compensation of their managers. This paper provides insights into the regulatory policy and the review framework for information disclosure in the context of registration system reform.